Grammarly, the popular writing-assistance platform, is facing a wave of criticism from academics after a new feature was found to include AI-generated versions of real scholars—including at least one who has died—as potential reviewers for manuscripts. The tool, called Expert Review, allows users to select an "expert" whose insights are used to generate feedback on their writing. The controversy erupted over the weekend when Verena Krebs, a medieval historian at Ruhr-University Bochum, shared a screenshot on social media showing historian David Abulafia listed among the available experts. Abulafia, a renowned scholar of Mediterranean history, passed away in January, making his inclusion particularly jarring.
The reaction was swift and sharp across academic circles. Vanessa Heggie, an associate professor of the history of science and medicine at the University of Birmingham, took to LinkedIn to voice her concerns. "Grammarly is now offering 'expert review' of your work by living and dead academics," she wrote. "Without anyone's explicit permission it's creating little LLMs based on their scraped work and using their names and reputation." Her post echoed a sentiment that quickly spread: the tool, while technologically impressive, raises serious ethical questions about consent and the use of individuals' intellectual labor.
Claire E. Aubin, a historian and host of the podcast "This Guy Sucked," called the feature "among the most cursed" things she had seen in academia, in a post on Bluesky that went viral. Kathleen Alves, an associate professor of English at CUNY, described it as "literally digital necromancy," a phrase that resonated with many. Hisham Zerriffi, an associate professor of forest resources management at the University of British Columbia, added a wry comment: "NecromancerLLM. Seriously, dead or alive, this is just wrong."
Grammarly's Expert Review is part of a suite of AI tools the company introduced last summer. According to its website, the feature is designed to help users "meet the expectations of your discipline and your project by drawing on insights from subject-matter experts and trusted publications." Users can open a document in Grammarly's AI platform, select the Expert Review agent, and receive suggestions based on the chosen expert. The tool can even generate revised versions of the text, with the option to "revise the draft yourself or let Expert Review rework things for you," as the site puts it.
Concerns Over Consent and Copyright
The backlash highlights broader anxieties about how AI models are built. Critics argue that the tool effectively impersonates real academics, using their names and reputations to lend credibility to AI-generated feedback. The fact that some of these scholars are deceased adds an extra layer of discomfort, as they cannot consent to or object to their inclusion. "This is literally digital necromancy," Alves reiterated, underscoring the sense of violation felt by many in the academic community.
Beyond the Expert Review feature, Grammarly also offers an "AI grader agent" that provides students with personalized feedback on homework by looking up "publicly available instructor information" about their teachers and professors. This has raised similar questions about the use of educators' professional identities without explicit permission.
The controversy comes at a time when the use of AI in academic settings is under increasing scrutiny. While tools like Grammarly have long been used for grammar and style checks, the introduction of AI agents that simulate specific individuals marks a significant shift. It raises fundamental questions about authorship, attribution, and the ethical boundaries of AI-generated content. As one academic put it, the feature feels invasive, not just for the living but also for the dead.
Grammarly has not yet publicly responded to the criticism. The company, which has positioned itself as a leader in AI-assisted writing, may need to address these concerns to maintain trust among its academic users. For now, the debate continues, with many scholars calling for greater transparency and consent mechanisms in the development of such tools.